How does the migration from another scheduler to IWS look like
Comarch’s experts have extensive experience in the automation of very large processing centers, by generating all IWS objects and importing them into IWS. Due to the number of objects (for example, more than 600,000 defined jobs) it was not possible to manually prepare processing in IWS. IWS objects were generated on the basis of job and relationship patterns, as well as configuration of running parameters. Preparation of tools for processing objects exported from CA Scheduler to objects imported to IBM Workload Scheduler seems to be a simpler task than preparing the above-mentioned IWS object generators.
Generally, to automate processing in IBM Workload Scheduler, regardless of whether it is already automated in another tool or is currently being manually started, you must specify IWS objects that will implement business and system processes, as follows:
| IWS Object | Characteristics | Description |
| IWS Application | Implemented by launching up to 255 IWS operations (jobs) in a defined order. | Implements a defined business process, or its logical fragment, or system work such as db2 utility |
| IWS application of type: Group | Type of IWS application with a defined start-up cycle (without substantive operations) | Groups IWS Applications that perform substantive or administrative tasks, run in the same way, i.e. having the same run cycle. |
| IWS operations | These can be jobs or started tasks sent to spool. They maybe virtual operations responsible for maintaining relationships. | The jobs or started task templates contain parameters that will be read and substituted based on variables tables, and templates for predefined actions, such as automatic restart in the event of a given error. They can also be virtual operations responsible for maintaining relationships. |
| Variables Tables | Contain variables that are dynamically read and inserted into runtime tasks when starting operations. | Variables in variables tables can be dynamically changed during system operation, for example, the number of threads of applications started may vary during the day, at night or during the weekend. You can run a process in IWS that, for example, changes variable values in variables tables every hour. |
| Calendars | Start-up calendars, necessary to define start-up cycles. | Business processes are run cyclically, cycles are different for different processes (for example, daily, only on business days, or only during weekends). Define calendars that provide the ability to define a launch cycle for each process. |
| Workstation | Where processing is to take place. | Certainly, you will need three types at each processing center: for jobs and started tasks that are automatically started, for manually started jobs, and for virtual operations. |
| Run Cycles | When applications are to be started. | Make the most of IWS Groups to define one run cycle for multiple applications. |
Depending on the processing characteristics in a given center, you need to decide which objects should be defined by programs generating objects that are then imported into IWS. For example, if you need to define two calendars (all business days and business days and holidays), it will be faster to define them manually. If all processing in a given center takes place in a relatively uniform cycle, you can define several groups manually and define run-time cycles for them.
Specify which business services are most important, which have an SLA or are critical. For such services, thorough analysis should be carried out in order to properly define duration and priority for operations included in the service, and the deadline for IWS applications implementing this process.
For many objects, mainly IWS operations, but often also IWS applications and variables tables, you must define job templates and object templates imported into IWS, and then prepare programs generating, based on the configuration, substantive tasks and objects to import into IWS.
In the whole system there is often a very large number of relationships between launched tasks. Here also you should usually define dependency templates and include them in the process of generating IWS objects.
Programs for generating objects should be prepared in such a way that they give the possibility of easily making changes in the processing process.
Piotr Kaszper
